2005-07-02 17:53:01 +04:00
|
|
|
#
|
2005-11-05 22:57:48 +03:00
|
|
|
# Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
|
|
|
|
|
# University Research and Technology
|
|
|
|
|
# Corporation. All rights reserved.
|
- Split the datatype engine into two parts: an MPI specific part in
OMPI
and a language agnostic part in OPAL. The convertor is completely
moved into OPAL. This offers several benefits as described in RFC
http://www.open-mpi.org/community/lists/devel/2009/07/6387.php
namely:
- Fewer basic types (int* and float* types, boolean and wchar
- Fixing naming scheme to ompi-nomenclature.
- Usability outside of the ompi-layer.
- Due to the fixed nature of simple opal types, their information is
completely
known at compile time and therefore constified
- With fewer datatypes (22), the actual sizes of bit-field types may be
reduced
from 64 to 32 bits, allowing reorganizing the opal_datatype
structure, eliminating holes and keeping data required in convertor
(upon send/recv) in one cacheline...
This has implications to the convertor-datastructure and other parts
of the code.
- Several performance tests have been run, the netpipe latency does not
change with
this patch on Linux/x86-64 on the smoky cluster.
- Extensive tests have been done to verify correctness (no new
regressions) using:
1. mpi_test_suite on linux/x86-64 using clean ompi-trunk and
ompi-ddt:
a. running both trunk and ompi-ddt resulted in no differences
(except for MPI_SHORT_INT and MPI_TYPE_MIX_LB_UB do now run
correctly).
b. with --enable-memchecker and running under valgrind (one buglet
when run with static found in test-suite, commited)
2. ibm testsuite on linux/x86-64 using clean ompi-trunk and ompi-ddt:
all passed (except for the dynamic/ tests failed!! as trunk/MTT)
3. compilation and usage of HDF5 tests on Jaguar using PGI and
PathScale compilers.
4. compilation and usage on Scicortex.
- Please note, that for the heterogeneous case, (-m32 compiled
binaries/ompi), neither
ompi-trunk, nor ompi-ddt branch would successfully launch.
This commit was SVN r21641.
2009-07-13 08:56:31 +04:00
|
|
|
# Copyright (c) 2004-2009 The University of Tennessee and The University
|
2005-11-05 22:57:48 +03:00
|
|
|
# of Tennessee Research Foundation. All rights
|
|
|
|
|
# reserved.
|
2009-04-24 20:39:33 +04:00
|
|
|
# Copyright (c) 2004-2009 High Performance Computing Center Stuttgart,
|
2005-07-02 17:53:01 +04:00
|
|
|
# University of Stuttgart. All rights reserved.
|
|
|
|
|
# Copyright (c) 2004-2005 The Regents of the University of California.
|
|
|
|
|
# All rights reserved.
|
2015-02-21 15:42:55 +03:00
|
|
|
# Copyright (c) 2009-2015 Cisco Systems, Inc. All rights reserved.
|
2005-07-02 17:53:01 +04:00
|
|
|
# $COPYRIGHT$
|
|
|
|
|
#
|
|
|
|
|
# Additional copyrights may follow
|
|
|
|
|
#
|
|
|
|
|
# $HEADER$
|
|
|
|
|
#
|
|
|
|
|
|
|
|
|
|
SUBDIRS = \
|
2006-02-12 04:33:29 +03:00
|
|
|
include \
|
2005-07-02 17:53:01 +04:00
|
|
|
asm \
|
- Split the datatype engine into two parts: an MPI specific part in
OMPI
and a language agnostic part in OPAL. The convertor is completely
moved into OPAL. This offers several benefits as described in RFC
http://www.open-mpi.org/community/lists/devel/2009/07/6387.php
namely:
- Fewer basic types (int* and float* types, boolean and wchar
- Fixing naming scheme to ompi-nomenclature.
- Usability outside of the ompi-layer.
- Due to the fixed nature of simple opal types, their information is
completely
known at compile time and therefore constified
- With fewer datatypes (22), the actual sizes of bit-field types may be
reduced
from 64 to 32 bits, allowing reorganizing the opal_datatype
structure, eliminating holes and keeping data required in convertor
(upon send/recv) in one cacheline...
This has implications to the convertor-datastructure and other parts
of the code.
- Several performance tests have been run, the netpipe latency does not
change with
this patch on Linux/x86-64 on the smoky cluster.
- Extensive tests have been done to verify correctness (no new
regressions) using:
1. mpi_test_suite on linux/x86-64 using clean ompi-trunk and
ompi-ddt:
a. running both trunk and ompi-ddt resulted in no differences
(except for MPI_SHORT_INT and MPI_TYPE_MIX_LB_UB do now run
correctly).
b. with --enable-memchecker and running under valgrind (one buglet
when run with static found in test-suite, commited)
2. ibm testsuite on linux/x86-64 using clean ompi-trunk and ompi-ddt:
all passed (except for the dynamic/ tests failed!! as trunk/MTT)
3. compilation and usage of HDF5 tests on Jaguar using PGI and
PathScale compilers.
4. compilation and usage on Scicortex.
- Please note, that for the heterogeneous case, (-m32 compiled
binaries/ompi), neither
ompi-trunk, nor ompi-ddt branch would successfully launch.
This commit was SVN r21641.
2009-07-13 08:56:31 +04:00
|
|
|
datatype \
|
|
|
|
|
etc \
|
2005-08-13 18:00:56 +04:00
|
|
|
util \
|
2005-10-17 04:21:10 +04:00
|
|
|
mca/base \
|
|
|
|
|
$(MCA_opal_FRAMEWORKS_SUBDIRS) \
|
|
|
|
|
$(MCA_opal_FRAMEWORK_COMPONENT_STATIC_SUBDIRS) \
|
2005-08-13 18:00:56 +04:00
|
|
|
. \
|
2005-10-17 04:21:10 +04:00
|
|
|
$(MCA_opal_FRAMEWORK_COMPONENT_DSO_SUBDIRS)
|
|
|
|
|
|
2011-04-08 18:59:49 +04:00
|
|
|
# libltdl is included by variable because if --disable-dlopen was
|
|
|
|
|
# used, there will be no generated Makefile in that directory (and
|
|
|
|
|
# therefore make distclean will fail).
|
2005-10-17 04:21:10 +04:00
|
|
|
DIST_SUBDIRS = \
|
2006-02-12 10:06:38 +03:00
|
|
|
include \
|
2005-10-17 04:21:10 +04:00
|
|
|
asm \
|
- Split the datatype engine into two parts: an MPI specific part in
OMPI
and a language agnostic part in OPAL. The convertor is completely
moved into OPAL. This offers several benefits as described in RFC
http://www.open-mpi.org/community/lists/devel/2009/07/6387.php
namely:
- Fewer basic types (int* and float* types, boolean and wchar
- Fixing naming scheme to ompi-nomenclature.
- Usability outside of the ompi-layer.
- Due to the fixed nature of simple opal types, their information is
completely
known at compile time and therefore constified
- With fewer datatypes (22), the actual sizes of bit-field types may be
reduced
from 64 to 32 bits, allowing reorganizing the opal_datatype
structure, eliminating holes and keeping data required in convertor
(upon send/recv) in one cacheline...
This has implications to the convertor-datastructure and other parts
of the code.
- Several performance tests have been run, the netpipe latency does not
change with
this patch on Linux/x86-64 on the smoky cluster.
- Extensive tests have been done to verify correctness (no new
regressions) using:
1. mpi_test_suite on linux/x86-64 using clean ompi-trunk and
ompi-ddt:
a. running both trunk and ompi-ddt resulted in no differences
(except for MPI_SHORT_INT and MPI_TYPE_MIX_LB_UB do now run
correctly).
b. with --enable-memchecker and running under valgrind (one buglet
when run with static found in test-suite, commited)
2. ibm testsuite on linux/x86-64 using clean ompi-trunk and ompi-ddt:
all passed (except for the dynamic/ tests failed!! as trunk/MTT)
3. compilation and usage of HDF5 tests on Jaguar using PGI and
PathScale compilers.
4. compilation and usage on Scicortex.
- Please note, that for the heterogeneous case, (-m32 compiled
binaries/ompi), neither
ompi-trunk, nor ompi-ddt branch would successfully launch.
This commit was SVN r21641.
2009-07-13 08:56:31 +04:00
|
|
|
datatype \
|
|
|
|
|
etc \
|
2005-10-17 04:21:10 +04:00
|
|
|
util \
|
|
|
|
|
mca/base \
|
|
|
|
|
$(MCA_opal_FRAMEWORKS_SUBDIRS) \
|
|
|
|
|
$(MCA_opal_FRAMEWORK_COMPONENT_ALL_SUBDIRS)
|
2005-07-02 17:53:01 +04:00
|
|
|
|
|
|
|
|
# Build the main OPAL library
|
|
|
|
|
|
configury: new OPAL_SET_LIB_PREFIX/ORTE_SET_LIB_PREFIX macros
These two macros set the prefix for the OPAL and ORTE libraries,
respectively. Specifically, the OPAL library will be named
libPREFIXopen-pal.la and the ORTE library will be named
libPREFIXopen-rte.la.
These macros must be called, even if the prefix argument is empty.
The intent is that Open MPI will call these macros with an empty
prefix, but other projects (such as ORCM) will call these macros with
a non-empty prefix. For example, ORCM libraries can be named
liborcm-open-pal.la and liborcm-open-rte.la.
This scheme is necessary to allow running Open MPI applications under
systems that use their own versions of ORTE and OPAL. For example,
when running MPI applications under ORTE, if the ORTE and OPAL
libraries between OMPI and ORCM are not identical (which, because they
are released at different times, are likely to be different), we need
to ensure that the OMPI applications link against their ORTE and OPAL
libraries, but the ORCM executables link against their ORTE and OPAL
libraries.
2014-10-22 16:49:58 +04:00
|
|
|
lib_LTLIBRARIES = lib@OPAL_LIB_PREFIX@open-pal.la
|
|
|
|
|
lib@OPAL_LIB_PREFIX@open_pal_la_SOURCES =
|
|
|
|
|
lib@OPAL_LIB_PREFIX@open_pal_la_LIBADD = \
|
2005-07-02 17:53:01 +04:00
|
|
|
asm/libasm.la \
|
- Split the datatype engine into two parts: an MPI specific part in
OMPI
and a language agnostic part in OPAL. The convertor is completely
moved into OPAL. This offers several benefits as described in RFC
http://www.open-mpi.org/community/lists/devel/2009/07/6387.php
namely:
- Fewer basic types (int* and float* types, boolean and wchar
- Fixing naming scheme to ompi-nomenclature.
- Usability outside of the ompi-layer.
- Due to the fixed nature of simple opal types, their information is
completely
known at compile time and therefore constified
- With fewer datatypes (22), the actual sizes of bit-field types may be
reduced
from 64 to 32 bits, allowing reorganizing the opal_datatype
structure, eliminating holes and keeping data required in convertor
(upon send/recv) in one cacheline...
This has implications to the convertor-datastructure and other parts
of the code.
- Several performance tests have been run, the netpipe latency does not
change with
this patch on Linux/x86-64 on the smoky cluster.
- Extensive tests have been done to verify correctness (no new
regressions) using:
1. mpi_test_suite on linux/x86-64 using clean ompi-trunk and
ompi-ddt:
a. running both trunk and ompi-ddt resulted in no differences
(except for MPI_SHORT_INT and MPI_TYPE_MIX_LB_UB do now run
correctly).
b. with --enable-memchecker and running under valgrind (one buglet
when run with static found in test-suite, commited)
2. ibm testsuite on linux/x86-64 using clean ompi-trunk and ompi-ddt:
all passed (except for the dynamic/ tests failed!! as trunk/MTT)
3. compilation and usage of HDF5 tests on Jaguar using PGI and
PathScale compilers.
4. compilation and usage on Scicortex.
- Please note, that for the heterogeneous case, (-m32 compiled
binaries/ompi), neither
ompi-trunk, nor ompi-ddt branch would successfully launch.
This commit was SVN r21641.
2009-07-13 08:56:31 +04:00
|
|
|
datatype/libdatatype.la \
|
2005-07-02 17:53:01 +04:00
|
|
|
mca/base/libmca_base.la \
|
2005-08-14 00:19:24 +04:00
|
|
|
util/libopalutil.la \
|
|
|
|
|
$(MCA_opal_FRAMEWORK_LIBS)
|
configury: new OPAL_SET_LIB_PREFIX/ORTE_SET_LIB_PREFIX macros
These two macros set the prefix for the OPAL and ORTE libraries,
respectively. Specifically, the OPAL library will be named
libPREFIXopen-pal.la and the ORTE library will be named
libPREFIXopen-rte.la.
These macros must be called, even if the prefix argument is empty.
The intent is that Open MPI will call these macros with an empty
prefix, but other projects (such as ORCM) will call these macros with
a non-empty prefix. For example, ORCM libraries can be named
liborcm-open-pal.la and liborcm-open-rte.la.
This scheme is necessary to allow running Open MPI applications under
systems that use their own versions of ORTE and OPAL. For example,
when running MPI applications under ORTE, if the ORTE and OPAL
libraries between OMPI and ORCM are not identical (which, because they
are released at different times, are likely to be different), we need
to ensure that the OMPI applications link against their ORTE and OPAL
libraries, but the ORCM executables link against their ORTE and OPAL
libraries.
2014-10-22 16:49:58 +04:00
|
|
|
lib@OPAL_LIB_PREFIX@open_pal_la_DEPENDENCIES = $(lib@OPAL_LIB_PREFIX@open_pal_la_LIBADD)
|
2015-04-15 00:34:21 +03:00
|
|
|
|
|
|
|
|
if HAVE_ATTRIBUTE_DESTRUCTOR
|
|
|
|
|
|
configury: new OPAL_SET_LIB_PREFIX/ORTE_SET_LIB_PREFIX macros
These two macros set the prefix for the OPAL and ORTE libraries,
respectively. Specifically, the OPAL library will be named
libPREFIXopen-pal.la and the ORTE library will be named
libPREFIXopen-rte.la.
These macros must be called, even if the prefix argument is empty.
The intent is that Open MPI will call these macros with an empty
prefix, but other projects (such as ORCM) will call these macros with
a non-empty prefix. For example, ORCM libraries can be named
liborcm-open-pal.la and liborcm-open-rte.la.
This scheme is necessary to allow running Open MPI applications under
systems that use their own versions of ORTE and OPAL. For example,
when running MPI applications under ORTE, if the ORTE and OPAL
libraries between OMPI and ORCM are not identical (which, because they
are released at different times, are likely to be different), we need
to ensure that the OMPI applications link against their ORTE and OPAL
libraries, but the ORCM executables link against their ORTE and OPAL
libraries.
2014-10-22 16:49:58 +04:00
|
|
|
lib@OPAL_LIB_PREFIX@open_pal_la_LDFLAGS = -version-info $(libopen_pal_so_version)
|
2005-07-02 17:53:01 +04:00
|
|
|
|
2015-04-15 00:34:21 +03:00
|
|
|
else
|
|
|
|
|
if HAVE_LD_FINI
|
|
|
|
|
|
|
|
|
|
lib@OPAL_LIB_PREFIX@open_pal_la_LDFLAGS = -version-info $(libopen_pal_so_version) -Wl,-fini,opal_cleanup
|
|
|
|
|
|
|
|
|
|
endif
|
|
|
|
|
endif
|
|
|
|
|
|
2005-10-17 04:21:10 +04:00
|
|
|
# included subdirectory Makefile.am's and appended-to variables
|
|
|
|
|
headers =
|
|
|
|
|
noinst_LTLIBRARIES =
|
2014-05-08 18:32:24 +04:00
|
|
|
dist_opaldata_DATA =
|
configury: new OPAL_SET_LIB_PREFIX/ORTE_SET_LIB_PREFIX macros
These two macros set the prefix for the OPAL and ORTE libraries,
respectively. Specifically, the OPAL library will be named
libPREFIXopen-pal.la and the ORTE library will be named
libPREFIXopen-rte.la.
These macros must be called, even if the prefix argument is empty.
The intent is that Open MPI will call these macros with an empty
prefix, but other projects (such as ORCM) will call these macros with
a non-empty prefix. For example, ORCM libraries can be named
liborcm-open-pal.la and liborcm-open-rte.la.
This scheme is necessary to allow running Open MPI applications under
systems that use their own versions of ORTE and OPAL. For example,
when running MPI applications under ORTE, if the ORTE and OPAL
libraries between OMPI and ORCM are not identical (which, because they
are released at different times, are likely to be different), we need
to ensure that the OMPI applications link against their ORTE and OPAL
libraries, but the ORCM executables link against their ORTE and OPAL
libraries.
2014-10-22 16:49:58 +04:00
|
|
|
lib@OPAL_LIB_PREFIX@open_pal_la_SOURCES += $(headers)
|
2005-10-17 04:21:10 +04:00
|
|
|
|
2005-07-02 17:53:01 +04:00
|
|
|
# Conditionally install the header files
|
|
|
|
|
|
2005-10-17 04:21:10 +04:00
|
|
|
if WANT_INSTALL_HEADERS
|
2014-05-08 06:01:35 +04:00
|
|
|
opaldir = $(opalincludedir)/$(subdir)
|
2010-09-25 02:53:28 +04:00
|
|
|
nobase_opal_HEADERS = $(headers)
|
2005-07-02 17:53:01 +04:00
|
|
|
endif
|
2005-10-17 04:21:10 +04:00
|
|
|
|
|
|
|
|
include class/Makefile.am
|
2005-11-11 03:26:27 +03:00
|
|
|
include memoryhooks/Makefile.am
|
2005-10-17 04:21:10 +04:00
|
|
|
include runtime/Makefile.am
|
|
|
|
|
include threads/Makefile.am
|
|
|
|
|
include mca/Makefile.am
|
2006-01-16 04:48:03 +03:00
|
|
|
include tools/Makefile.am
|
2008-02-28 04:57:57 +03:00
|
|
|
include dss/Makefile.am
|